OpenGVLab/InternVL3-2B-Instruct
Full fine-tune vs OpenGVLab/InternVL3-2B-Pretrained. Fine-tune: 100% of the weights changed, across 28 of 28 layers.
Tensors changed: lm_head 1/1, Embeddings 5/5, Norms 141/153, MLP 186/186, Attention 292/292, other 48/48
6 derivatives checked
- OpenGVLab/InternVL3-2B: Full fine-tune
- OpenGVLab/InternVL3-2B-hf: Different architecture
- ggml-org/InternVL3-2B-Instruct-GGUF: Same weights, converted (fewer layers)
- unsloth/InternVL3-2B-Instruct-GGUF: Same weights, converted (fewer layers)
- mradermacher/InternVL3-2B-Instruct-i1-GGUF: Quantization (fewer layers)
- mradermacher/InternVL3-2B-Instruct-GGUF: Same weights, converted (fewer layers)
2 other repos hold byte-for-byte identical weight files.